An interval [0, 1] of the output of neural network is equally divided into M (M being an integer of two or more), and the numbers or frequencies of data for the correct/incorrect patterns contained in the i-th interval [(i - 1)/M, i/M] are µ1i and µ0i, respectively (where, i = 1 ... M). In this case, if this network provides an output contained in the i-th interval to unknown pattern data, this pattern is stored as a likelihood conversion table so that the pattern outputs likelihood P1i, which is a category, in an equation P1i = (µ1i + 1)/(µ1i + µ0i + 2). Then, when a value contained in the i-th interval [(i - 1)/M, i/M] is output from a neural network, the likelihood convertor receives it as an input and outputs P1i which is so to speak normalized likelihood.

Because the intervention of such likelihood convertor makes it possible for each network to be adjusted to output likelihood, the variation of learning is corrected for the networks so that the output from each network can be compared with each other. That is, a category provided the highest likelihood can be determined to be the most probable candidate.